Wearable smart devices are now playing an important role in modern society. They can be used for locating the user thanks to embedded GNSS receivers and/or inertial sensors. A method based on GNSS carrier phase measurements, i.e. Time Difference Carrier Phase (TDCP), has shown a good performance in estimating user's velocity. However, in complex environments such as urban area where a large amount of multi-path reception could take place, the velocity estimated by TDCP could still be distorted. To inform the vulnerable users of the uncertainty on the velocity estimation for the safety-critical applications, a statistical velocity error bound, namely velocity Protection Level (PL), is calculated in this paper. In compliance with a pre-defined maximum velocity error tolerance called Alert Limit (AL), the velocity PL should be below or equal to the velocity AL. Otherwise, the user must be alerted in order to prevent him/her from dangerous accidents. In this research work, we proposed a horizontal velocity PL calculation method which is based on TDCP and Kalman filter, for the application of wearable positioning devices. A velocity AL is proposed for a specific scenario of blind people with white cane, since there is not any existing integrity specifications for pedestrian navigation. In urban environment, the velocity estimation mean error with proposed method is between 0.036 m/s and 0.053 m/s, and the Hazardously Misleading Information (HMI) rate can achieve the level of 10?3, for a specific use case of blind people with a white cane.